Page 51 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 3 – Internet of Bio-Nano Things for health applications

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 3

culated by integrating the propensity function of concen‑
Wearable device with tration signal intensity over a symbol duration. Finally,
transmitter and receiver
at the receiver, the threshold‑based detection was per‑
Tx Rx formed where the test statistic was shown to be com‑
2
Molecules Skin posed of a constant, two Normal random variables, and a
transmission Epidermis Signaling Chi‑square random variable. Further, in [36], binary and
through a needle Derma 3‑ary PAM transmission schemes have been used. In this
Blood vessel work, the detection at the receiver was carried out based
Blood flow
direction on the response generated by the bacteria to the input
Red blood cell molecular signals. Four different sampling strategies of
the response were studied. These strategies were based
on the total response, peak response, positive and nega‑
Fig. 6 – Wearable device for monitoring blood viscosity.
tive slopes in which the BER for positive slope sampling
Diseased cell Cell was found to be the worst.
surface surface
Tx Rx receptor
In [37], the performance of NP and matched ilter (MF)
detectors were analyzed for the detection of a nano‑
Nucleus biosensor signal corrupted with thermal and shot noises.
It is shown that these two detectors result in identical per‑
formance in the presence of thermal noise only. However,
as the shot noise increases, the NP‑based test outperforms
the MF detector. Also, when the binary signals to be dif‑
Signaling Drug ferentiated, are very close or farther from each other, both
molecule molecule
NP and MF detectors have identical performance other‑
wise the NP rule outperforms. Further, binary and quater‑
Fig. 7 – Drug delivery over a cell surface using MIMO‑MC system.
nary ASK were used as the transmission schemes in [38]
A general block diagram of an MC system proposed in [33] and the receiver based on an NP test was proposed. In ad‑
has been shown in Fig. 10, where y(t) is the received sig‑ dition to this, suboptimal detection schemes with reduced
nal, N tx is the number of transmitted molecules, p(t) is the complexity were also reported.
pulse shaping ilter, h(t) is the CIR, and n(t) is the signal‑
On the other hand, OOK modulation and two different
dependent noise (other external noise sources can also be
methods of detection based on pulse amplitude and pulse
included). The received number of molecules is assumed
energy with ixed threshold have been proposed in [39].
to be binomial which is approximated as a Gaussian dis‑
In this work, analytical expressions of communication
tributed random variable to derive the maximum likeli‑
metrics such as pulse delay (i.e., peak time), pulse am‑
hood detector. In addition to the maximum likelihood‑
plitude (i.e., the concentration at peak time), and pulse
based detector, a sequence detection scheme (i.e., Viterbi
width (i.e., the time difference between two points when
detector) is also derived since the channel with ISI adds
concentration falls to 50% of its peak value) were de‑
memory to the signal which is similar to a convolutional
rived for the pulse amplitude‑based detector. Similarly,
encoder using the shift registers.
the analytical expressions of pulse energy (i.e., the inte‑
For improving the BER performance compared to [33], gral of concentration with respect to time) and pulse du‑
the authors in [34] introduced a noise whitening ilter at ration (i.e., the time when the energy becomes a speci‑
the receiver for binary MoSK modulation‑based transmis‑ ied fraction of total energy) have been derived for a pulse
sion. In this work, the reception process was modeled as a energy‑based detector. It is observed therein that the
Low Pass Filtering (LPF) operation where the LPF trans‑ pulse amplitude and energy‑based detectors are suitable
fer function was derived using the ligand‑receptor bind‑ for a high transmission rate and large transmission dis‑
ing equation at the receiver. Moreover, the Brownian and tance, respectively.
residual noises corrupting the signal have been assumed
In [40] the transmitter used a rectangular pulse of con‑
to be signal‑dependent. Finally, the system bit error rate
centration. Four different methods of detection have been
performance is shown to improve by using a noise whiten‑
ing ilter. proposed at the receiver. More speci ically, Maximum A
posteriori Probability (MAP) and maximum likelihood‑
The work in [35] derived an energy‑based detector con‑ based sequence detection schemes were proposed, which
sidering the binary Pulse Amplitude Modulation (PAM) maximize the joint Probability Density Function (PDF) of
based transmission using a square pulse instead of an im‑ received samples and the transmitted bits, and use the
pulse of molecules. For this setup, the mean and vari‑ Viterbi algorithm. Let b j denote the information bit trans‑
ance of the number of received molecules have been cal‑
2 The test statistic is derived by taking the logarithm of the likelihood
ratio that uses the Neyman‑Pearson (NP) formula.

© International Telecommunication Union, 2021 39